HiSIM2 MOSFET Model Extraction

opt_ex16 : HiSIM2 MOSFET Model Extraction

Requires: Utmost IV, SmartSpice, SmartView

Minimum Versions: Utmost IV 1.11.2.R, SmartSpice 4.16.1.R, SmartView 2.30.3.R

This example describes how to extract a standard HiSIM2 model. To extract a model which is scalable with geometry, multiple different device sizes must be included. In this example, twelve different geometry devices for DC and two geometry devices for capacitance at room temperature are included.

The project file opt_ex16.prj and the data file opt_ex16.uds for this example should be loaded into your database. When opened, the project will look as shown in opt_ex16_project.png .

The optimization sequence, which automates the parameter extraction, is divided into twelve sections. Some model parameters are optimized in multiple sections for better fitting.

Before starting the optimization sequence, the following process parameters are defined in the model library. The oxide thickness, tox, is set at this point. The other parameter values are used as starting values for further optimization during the sequence.

  • tox:Physical oxide thickness
  • nsubc:Substrate-impurity concentration
  • nsubp:Maximum pocket concentration
  • lover:Overlap length
  • nover:Impurity concentration in overlap region

Section 1 : IdVglin_large_1_rt

This section optimizes the parameters for linear gate turn on characteristics at large size device. The data is the drain current vs. gate voltage characteristics at low drain voltage (IDvsVG_lin). This section has two steps. The first step uses the Genetic Algorithm(GA) optimizer and the second step uses the Levenberg-Marquardt Algorithm(LM) optimizer. The following parameters are optimized;

  • nsubc:Substrate-impurity concentration
  • vfbc:Flat-band voltage
  • ndep:Depletion charge contribution on effective-electric field
  • muecb0:Coulomb scattering
  • muecb1:Coulomb scattering
  • mueph0:Phonon scattering
  • mueph1:Phonon scattering

After this section is completed, the fit to measured data is as shown in opt_ex16_01.png .

Section 2 : Cgg_large_rt

This section optimizes the parameters for total gate capacitance (Cgg) characteristics. The data is Cgg and algorithm is LM. The following parameters are optimized;

  • lover:Overlap length
  • vfbover:Flat-band voltage in overlap region
  • pgd1:Strength of poly-depletion effect
  • pgd2:Threshold voltage of poly-depletion effect

After this section is completed, the fit to measured data is shown in opt_ex16_02.png .

Section 3 : IdVglin_large_2_rt

This section optimizes the parameters for threshold, mobility and resistance characteristics at large size device. This section has three steps and all steps use LM. The following parameters are optimized;

  • muesr0:Surface-roughness scattering
  • muesr1:Surface-roughness scattering
  • ninv:Inversion charge contribution on effective-electric field

The following model parameters, which have already been optimized previously, are again optimized to ensure the fit to the entire data.

  • nsubc
  • vfbc
  • ndep
  • muecb0
  • muecb1
  • mueph0
  • mueph1

After this section is completed, the fit to measured data is shown in opt_ex16_03.png .

Section 4 : Cover_all_rt

This section optimizes the parameters for overlap capacitance characteristics. This uses the Cgg, Cgd and Cgs data. This section has two steps, the first step uses GA and the second step uses LM. The following parameters are optimized;

  • xld:Gate-overlap length
  • xwd:Gate-overlap width
  • nover:Impurity concentration in overlap region
  • pgd4:L dependence of poly-depletion effect
  • cgdo:Gate-to-drain overlap capacitance
  • cgso:Gate-to-source overlap capacitance

The following model parameters, which have already been optimized previously, are again optimized to ensure the fit to the entire data.

  • lover
  • vfbover

After this section is completed, the fit to measured data is shown in opt_ex16_04.png .

Section 5 : IdVglin_larray_rt

This section optimizes the parameters for threshold, mobility and resistance characteristics for devices with large width and various lengths. This section has two steps and all steps use LM. The following parameters are optimized;

  • nsubp:Maximum pocket concentration
  • lp:Pocket penetration length
  • muephl:Length dependence of phonon mobility reduction
  • muesrl:Length dependence of surface roughness mobility reduction
  • ndepl:Modification of QB contribution for short-channel case
  • parl2:Depletion width of channel/contact junction
  • sc1:Magnitude of short-channel effect
  • sc3:Vds dependence of short-channel effect

After this section is completed, the fit to measured data is shown in opt_ex16_05.png .

Section 6 : IdVglin_all_1_rt

This section optimizes the parameters for threshold, mobility and resistance characteristics for all device sizes. The data is IDvsVG_lin and this section has three steps, all of which use LM. The following parameters are optimized;

  • muephs:Mobility modification due to small size
  • muephw:Width dependence of phonon mobility reduction
  • muepld:Length dependence of phonon mobility reduction
  • muepwd:Width dependence of phonon mobility reduction
  • muesrw:Change of surface roughness related mobility
  • scp1:Magnitude of short-channel effect due to pocket
  • scp3:Vds dependence of short-channel due to pocket
  • wfc:Threshold voltage change due to capacitance change
  • wl2:Threshold voltage shift due to small size effect
  • wl2p:Threshold voltage shift due to small size effect
  • wvth0:Threshold voltage shift
  • ndepw:Modification of universal mobility for narrow width
  • ndepwp:Modification of universal mobility for narrow width

The following model parameters, which have already been optimized previously, are again optimized to ensure the fit to the entire data.

  • muephl
  • muesrl
  • ndep
  • ndepl
  • parl2
  • ninv
  • sc1
  • sc3
  • xld
  • xwd

After this section is completed, the fit to measured data is shown in opt_ex16_06.png .

Section 7 : IdVglin_IdVd_all_rt

This section optimizes the parameters for threshold and low gate voltage region characteristics at all device sizes. The data is IDvsVG_lin and the drain current vs. drain voltage characteristics at low body voltage (IDvsVD_low), and this section has three steps. The first and third steps use LM, the second step uses GA. The following parameters are optimized;

  • nsubcw:Modification of substrate concentration for narrow width
  • nsbcwp:Modification of substrate concentration for narrow width
  • nsubpl:Initial channel length of NSUBP reduction
  • nsubpw:Modification of pocket concentration for narrow width
  • nsubpwp:Modification of pocket concentration for narrow width
  • clm1:Hardness coefficient of channel/contact junction
  • clm2:Coefficient for QB contribution
  • clm3:Coefficient for QI contribution
  • clm5:Effect of pocket implantation
  • clm6:Effect of pocket implantation
  • gidl1:Magnitude of GIDL
  • gidl2:Field dependence of GIDL
  • gidl3:Vds dependence of GIDL
  • gidl4:Threshold of Vds dependence
  • gidl5:Correction of high-field contribution
  • gidl6:Threshold of Vbs dependence
  • gidl7:Correction of high-field contribution
  • glksd1:Gate to source/drain current coefficient
  • sc2:Vds dependence of short-channel effect
  • scp2:Vds dependence of short-channel due to pocket
  • sc4:Coupling of Vds-Vbs dependence
  • sub1:Substrate current coefficient of magnitude
  • sub1l:Lgate dependence SUB1
  • sub2:Substrate current coefficient of exponential term
  • svbs:Substrate current dependence on Vbs
  • svgs:Substrate current dependence on Vgs
  • svgsl:L dependence of SVGS
  • svgsw:W dependence of SVGS

The following model parameters, which have already been optimized previously, are again optimized to ensure the fit to the entire data.

  • nsubp
  • parl2
  • sc1
  • sc3
  • scp1
  • scp3
  • wfc
  • wl2
  • wl2p
  • wvth0

After this section is completed, the fit to measured data is shown in opt_ex16_07.png .

Section 8 : IdVglin_all_2_rt

The purpose of this section is to tune the parameters that were optimized in the previous sections. This section optimizes the parameters for the linear region characteristics at all device sizes. This section has three steps and all steps use LM. The following parameters are re-optimized;

  • mueph0
  • mueph1
  • muesr0
  • muesr1
  • muephl
  • muephs
  • muephw
  • muepld
  • muepwd
  • muesrl
  • muesrw
  • nsubcw
  • ndep
  • ndepl
  • ndepw
  • ndepwp
  • ninv
  • sc1
  • sc3
  • scp1
  • scp3
  • wfc
  • wl2
  • wl2p
  • wvth0

After this section is completed, the fit to measured data is shown in opt_ex16_08.png .

Section 9 : IdVdlow_rt

This section optimizes the parameters for saturation region characteristics at all device sizes. The data is IDvsVD_low. This section has three steps and all steps use LM. The following parameters are optimized;

  • ninvd:Reduced resistance effect for small Vds
  • mueslp:Length dependence of surface roughness mobility reduction
  • mueswp:Change of surface roughness related mobility
  • muepsp:Mobility modification due to small size
  • vmax:Saturation velocity
  • vover:Velocity overshoot effect
  • voverp:Leff dependence of velocity overshoot
  • vovers:Modification of maximum velocity due to small size
  • voversp:Modification of maximum velocity due to small size
  • svds:Substrate current dependence on Vds
  • sub2l:L dependence of SUB2
  • ndeplp:Modification of QB contribution for short-channel case
  • ndepw:Modification of universal mobility for narrow width
  • ndepwp:Modification of universal mobility for narrow width

The following model parameters, which have already been optimized previously, are again optimized to ensure the fit to the entire data.

  • clm1
  • clm2
  • clm3
  • clm5
  • clm6
  • muesr0
  • muesr1
  • muesrl
  • muesrw
  • muephs
  • wvth0
  • ninv
  • wl2
  • wl2p
  • ndep
  • ndepl
  • sc2
  • scp2
  • sub1
  • sub1l
  • sub2
  • svgs
  • svgsl
  • pgd1
  • pgd2
  • pgd4

After this section is completed, the fit to measured data is shown in opt_ex16_09.png .

Section 10 : IdVg_all_rt

This section optimizes the parameters for linear and saturation region characteristics at all device sizes. The data used are IDvsVG_lin and the drain current vs. gate voltage characteristics at high drain voltage (IDvsVG_sat) This section has one step which uses LM. The following parameters are optimized;

  • sc3vbs:For diminishing of impunity gradient
  • sc4:Coupling of Vds-Vbs dependence
  • bs1:Body-coefficient modification by impurity profile
  • bs2:Body-coefficient modification by impurity profile
  • bb:High-field-mobility degradation
  • svbs:Substrate current dependence on Vbs
  • svbsl:Lgate dependence of SVBS
  • svbslp:Lgate dependence of SVBS

The following model parameters, which have already been optimized previously, are again optimized to ensure the fit to the entire data.

  • muephl
  • muephs
  • muephw
  • muepld
  • muepwd
  • muesrl
  • muesrw
  • ndep
  • ndepl
  • ndepw
  • ndepwp
  • ninv
  • sc1
  • sc2
  • sc3
  • scp1
  • scp3
  • wfc
  • wl2
  • wl2p
  • wvth0

After this section is completed, the fit to measured data is shown in opt_ex16_10.png .

Section 11 : IdVd_all_rt

This section optimizes the parameters for low and high body voltage characteristics at all device sizes. The data used are IDvsVD_low and the drain current vs. drain voltage characteristics at high body voltage (IDvsVD_high). This section has one step which uses LM. The following parameters are optimized;

  • clm1
  • clm2
  • clm3
  • clm5
  • clm6
  • wvth0
  • ninv
  • ninvd
  • wl2
  • wl2p
  • ndep
  • ndepl
  • ndeplp
  • ndepw
  • ndepwp
  • vmax
  • vover
  • voverp
  • vovers
  • voversp
  • bs1
  • bs2
  • bb
  • svbs
  • svbsl
  • svbslp
  • pgd1
  • pgd2
  • pgd4

After this section is completed, the fit to measured data is shown in opt_ex16_11.png .

Section 12 : IdVg_IdVd_all_rt

The final section is for all geometry devices for the improvement of the entire fitting. All of the DC data is used, and the optimizer algorithm is LM. The following parameters are optimized;

  • nsubc
  • nsubp
  • vfbc
  • ndep
  • muesr0
  • muesr1
  • muesrl
  • muesrw
  • sc1
  • sc3
  • sc4
  • wfc
  • wl2
  • wl2p
  • wvth0
  • ninv
  • ninvd
  • mueslp
  • mueswp
  • vmax
  • vover
  • voverp
  • vovers
  • voversp
  • bb

After this section is completed, the fit to measured data is shown in opt_ex16_12.png .

When the sequence is completed, the final model card can be exported into an external model library file as shown in opt_ex16.lib.